Professional Documents
Culture Documents
3, JULY 2010
Abstract—Shunt capacitors installation in distribution systems systems supply single and three-phase loads through distribu-
requires optimal placement and sizing. More harmonics are being tion transformers. Second, the phases of transmission lines are
injected into distribution systems. Adding shunt capacitors may unequally loaded. Third, unlike those in transmission systems
lead to high distortion levels. The capacitor placement and sizing
problem is a nonlinear integer optimization problem, with loca- overhead lines in distribution systems are not transposed.
tions and ratings of shunt capacitors being discrete values. The goal Due to the widespread use of harmonic-producing equipment
is to minimize the overall cost of the total real power loss and that in distribution systems, harmonics are propagated throughout
of shunt capacitors while satisfying operating and power quality those systems. Harmonics are undesirable and cause equipment
constraints. This paper proposes to solve the problem using par- overheating due to the excessive losses and potential malfunc-
ticle swarm optimization (PSO). A discrete version of PSO is com-
bined with a radial distribution power flow algorithm (RDPF) to tioning of electric equipment. Inclusion of shunt capacitors
form a hybrid PSO algorithm (HPSO). The former is employed without considering the presence of harmonic sources in the
as a global optimizer to find the global optimal solution, while the system may lead to an increase in harmonic distortion levels
latter is used to calculate the objective function and to verify bus due to resonance between capacitors and the various inductive
voltage limits. To include the presence of harmonics, the devel- elements in the system.
oped HPSO was integrated with a harmonic power flow algorithm
(HPF). The proposed (HPSO-HPF)-based approach is tested on an Baghzouz developed a local variations-based heuristic ap-
IEEE 13-bus radial distribution system (13-Bus-RDS). The find- proach to find the global optimal ratings of shunt capacitors such
ings clearly demonstrate the necessity of including harmonics in that the cost of total real power loss and that of shunt capacitors
optimal capacitor placement and sizing to avoid any possible prob- were minimized [1]. The optimal capacitor sizing problem was
lems associated with harmonics. formulated as a nonlinear integer programming problem with
Index Terms—Harmonics, particle swarm, shunt capacitors. inequality constraints. The constraints considered were the rms
values of bus voltages and total harmonic distortions. The only
harmonic source assumed was the utility substation. A heuristic
I. INTRODUCTION
algorithm based on local variations was proposed to overcome
the prohibitive computational time associated with considering
while rms values of bus voltages and total harmonic distortions problem with both locations and sizes of shunt capacitors
being kept within allowable limits. Nikham et al. also used a being discrete. The objective function encompasses the total
genetic algorithm to solve the optimal capacitor allocation and cost of the total real power loss and that of shunt capacitors.
sizing problem taking the presence of harmonic sources into The objective function is restricted by equality and inequality
account [5]. The objective function consisted of the cost of real constraints.
power losses and that of shunt capacitors to be installed. The Objective Function: The goal is to minimize the cost of the
cost associated with the reactive power injection was fixed for total real power loss and that of the shunt capacitor installation.
all possible capacitor sizes. In other words, the cost of the re- The cost function is given by
active power injected was assumed to be constant independent
of the capacitor size. Masoum et al. developed a hybrid tool (1)
based on maximum sensitivity selection (MSS) and local vari-
ations (LV) to solve the optimal capacitor placement and sizing
problem [6]. The former was used to enhance the convergence where
speed by narrowing down the search space, while the latter was
annual cost per unit of the real power loss
employed to find the global optimal solution. Three harmonic
($/kW/year);
distortion levels were considered for the system investigated.
The system under investigation involved only one harmonic annual cost per unit of the reactive power injection
source and that was a six-plus converter. The results of the at bus i ($/kVAR/year);
hybrid MSS-LV algorithm were compared with those of the reactive power injection at bus i (kVAR);
MSS-based algorithm. In later work, Masoum et al. applied
a fuzzy logic-based algorithm to solve the same problem [7]. total number of shunt capacitors to be installed;
Both the objective function and the constraints were fuzzified. total real power loss (kW).
Alpha cuts were used to direct the search process and to ensure
that the objective function improved each time. The candidate The total real power loss is defined by
buses were determined according to the objective function,
constraints and reactive power compensation sensitivities. Two
harmonic distortion levels were considered this time to compare
the results obtained with those obtained by the MSS-based al-
gorithm. A conclusion was drawn that the appropriate locations (2)
and ratings of shunt capacitors would not only improve voltage where
profiles but also would reduce harmonic distortion levels.
Masousm et al. took advantage of the capability of genetic number of branches;
algorithms (GAs) to escape local optima [8]. Improvements smallest harmonic order of interest;
in voltage profiles and power quality were achieved through
the proper installation of fixed shunt capacitors in distorted highest harmonic order of interest.
distribution systems. The applicability of GA-based approach The fundamental component of the total real power loss is
was proven to yield to better results compared to the previous calculated using a three phase power flow algorithm (RDPF)
work done by the same authors. [11].
Another method based on particle swarm optimization (PSO) The harmonic component of the total real power loss is com-
was offered in [9] to solve the capacitor placement and sizing puted by a harmonic power flow algorithm (HPF) [12].
problem considering harmonics. The problem was mathemati- Note that the harmonic component of the total real power loss
cally modeled as a nonconvex optimization problem. The ob- is small compared with the fundamental one. However, this por-
jective function was augmented by quadratic penalty functions tion of the total real power loss increases as harmonic-producing
to account for inequality constraints. That is, the objective func- loads continue to increase in RDS. Consequently, the undesir-
tion was penalized whenever the inequality constraints were vi- able presence of harmonics will cause more equipment over-
olated. The proposed PSO algorithm did not account for unbal- heating, stress on equipment insulation, and equipment failure.
anced operating conditions. Khalil et al. [10] proposed a binary Not to mention of course the interference with communication
PSO algorithm to find the best locations and ratings of fixed networks. It should be pointed out that the cost of the real power
shunt capacitors in balanced distribution systems. The only har- loss per unit is fixed. However, the cost of the reactive power in-
monic source considered was the substation voltage. Their ob- jection per unit varies from one capacitor size to another [1].
jective was to properly place and size shunt capacitors while Generally, the larger the capacitor size is, the cheaper it be-
keeping the cost of real power losses and that of shunt capacitors comes.
at a minimum. The objective function was subject to equality Constraints: Along with the objective function, there is an-
and inequality constraints. other significant part of the optimization model that needs to
be defined and that is the constraints. In real applications, there
II. PROBLEM FORMULATION are always limits on the choices of control variables. The con-
The optimal capacitor placement and sizing problem is straints considered in this research are of two types: equality and
formulated as a constrained nonlinear integer optimization inequality.
1736 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 25, NO. 3, JULY 2010
A. Equality Constraints itors that are commercially available come in discrete sizes. That
The equality constraints are those associated with the non- is, the shunt capacitors to be dealt with are multiple integers of
linear power flow equations. It is noted in many published pa- the smallest capacitor size available
pers that the power flow equations are the real and reactive
(7)
power mismatch equations. The reason for this is that modified
versions of conventional power flow programs such as Newton- where
Raphson method and Gauss Siedel method are widely used. In
this work, the power flow representation is based on bus current smallest capacitor size available.
injections and thus the equality constraints are the bus current
It is worth mentioning that the total reactive power injection is
mismatch equations. The equality constraints are expressed in a
not to exceed the total reactive power demand in RDS
vector form as follows:
(3) (8)
where:
where
vector of state (dependent) variables;
total reactive power demand.
vector of control (independent) variables.
III. PARTICLE SWARM OPTIMIZATION
B. Inequality Constraints
PSO is a metaheuristic optimization technique developed in
The inequality constraints are those associated with the bus 1995 by Kennedy and Eberhart [13]–[15]. The fundamental idea
voltages, total harmonic distortion levels, and shunt capacitors behind the PSO algorithm is that a population called a swarm is
to be installed. randomly generated. The swarm consists of individuals called
1) Bus Voltage Limits: The bus voltage magnitudes are to particles. Each particle in the swarm represents a potential solu-
be kept within acceptable operating limits throughout the opti- tion of the optimization problem. Each particle moves through
mization process a D-dimensional search space at a random velocity. Each par-
ticle updates its velocity and position according to the following
(4)
equations:
where
inertia weight;
(5) acceleration constants;
two random numbers in the range of [0, 1];
where
best position ever visited by the particle at the
number of buses. th iteration;
The rms value of the th bus voltage involves only the funda- ;
mental component, the first term of (5), when harmonics are not global best position in the entire swarm;
of interest.
2) Total Harmonic Distortion Limits: The total harmonic dis- .
tortion at each bus is to be kept less or equal to the maximum
allowable harmonic distortion level as shown
IV. SOLUTION METHOD
(6) As one of this research objectives, a hybrid particle swam
optimization algorithm (HPSO) was developed to find the best
where combination of the locations and sizes of single phase shunt ca-
pacitors in unbalanced radial distribution systems (unbalanced-
maximum allowable harmonic distortion RDS). The HPSO algorithm combines a discrete version of PSO
level at each bus. with a three phase power flow algorithm (RDPF). The former
3) Number and Sizes of Shunt Capacitors: There are con- is employed as a global optimizer to optimally locate and rate
straints associated with the shunt capacitors themselves. Capac- shunt capacitors, while the latter is utilized to minimize the bus
EAJAL AND EL-HAWARY: OPTIMAL CAPACITOR PLACEMENT AND SIZING 1737
TABLE I TABLE II
COMMERCIALLY AVAILABLE CAPACITOR SIZES RESULTS OF THE OPTIMAL PLACEMENT AND SIZING OF ONE SHUNT CAPACITOR
WITH THEIR CORRESPONDING COSTS IN A 13-BUS-RDS
TABLE III
VOLTAGE PROFILES AND TOTAL HARMONIC DISTORTIONS OF A 13-BUS-RDS
FOR CASES 1, 2, AND 3 OF THE OPTIMAL PLACEMENT AND SIZING OF ONE
SHUNT CAPACITOR
TABLE IV
RESULTS OF THE OPTIMAL PLACEMENT AND SIZING OF
THREE SHUNT CAPACITORS IN A 13-BUS-RDS
TABLE V
VOLTAGE PROFILES AND TOTAL HARMONIC DISTORTIONS OF A 13-BUS-RDS
FOR CASES 1, 2, AND 3 OF THE OPTIMAL PLACEMENT AND SIZING OF THREE
SHUNT CAPACITORS
undesirable harmonic distortion levels causing more damage to [13] J. Kennedy and R. Eberhart, “Particle swarm optimization,” in
the electric equipment of both the electric utility and customers. Proc. IEEE Int. Conf. Neural Networks, Piscataway, NJ, 1995, pp.
1942–1948.
[14] R. Eberhart and J. Kennedy, “A new optimizer using particle swarm
ACKNOWLEDGMENT theory,” in Proc. 6th Int. Symp. Micro Machine and Human Science,
1995, pp. 39–43.
The authors would like to thank the reviewers for their valu- [15] X. Hu, Y. Shi, and R. Eberhart, “Recent advances in particle swarm,”
able comments and recommendations. in Proc. Congr. Evolutionary Computation, 2004, vol. 1, pp. 90–97.
[16] M. R. AlRashidi and M. E. El-Hawary, “Hybrid particle swarm opti-
REFERENCES mization approach for solving the discrete OPF problem considering
the valve loading effects,” IEEE Trans. Power Syst., vol. 22, no. 4, pp.
[1] Y. Baghzouz and S. Ertem, “Shunt capacitor sizing for radial distri- 2030–2038, Nov. 2007.
bution feeders with distorted substation voltages,” IEEE Trans. Power [17] W. H. Kersting, “Radial distribution test feeders,” IEEE Trans. Power
Del., vol. 5, no. 2, pp. 650–657, Apr. 1990. Syst., vol. 6, no. 3, pp. 975–985, Jul. 1991.
[2] Y. Yan, T. Wu, and C. Ho, “Optimal capacitor placement using sensi- [18] Task Force on Harmonics Modeling and Simulation, “Modeling and
tivity index and differential evolution for the three-phase distribution simulation of the propagation of harmonics in electric power networks
systems with harmonic distortion,” in Proc. IASTED Int. Conf. Power part II: Sample systems and examples,” IEEE Trans. Power Del., vol.
and Energy Systems, May 13–15, 2002, pp. 352–353. 11, no. 1, pp. 466–474, Jan. 1996.
[3] G. Carpinelli, P. Varilone, V. Di Vito, and A. Abur, “Capacitor place- [19] Y. Shi and R. Eberhart, “A modified particle swarm optimizer,” in Proc.
ment in three-phase distribution systems with nonlinear and unbal- IEEE Int. Conf. Evolutionary Computation, 1998, pp. 69–73.
anced loads,” Proc. Inst. Elect. Eng., Gen., Transm. Distrib., vol. 152,
no. 1, pp. 47–52, 2005. Abdelsalam A. Eajal (S’00) received the B.S. de-
[4] A. Abou-Ghazala, “Optimal capacitor placement in distribution sys- gree in electrical engineering from the University of
tems feeding nonlinear loads,” in Proc. IEEE Bologna Power Tech Altahadi, Sirt, Libya, in 2001 and the M.A.Sc. degree
Conf., 2003, pp. 97–103. in electrical engineering from Dalhousie University,
[5] T. Niknam, A. M. Ranjbar, H. Arabian, and M. Mirjafari, “Optimal Halifax, NS, Canada, in 2008.
reactive power planning in harmonic distorted power system using ge- Currently, he is a staff member at Altahadi Univer-
netic algorithm,” in Proc. IEEE Region 10 Conf., 2004, pp. 347–350. sity, Sirt, Libya. His research interests involve appli-
[6] M. A. S. Masoum, M. Ladjevardi, E. F. Fuchs, and W. M. Grady, “Ap- cations of optimization techniques in power system
plication of local variations and maximum sensitivities selection for planning and operation.
optimal placement of shunt capacitor banks under nonsinusoidal oper-
ating conditions,” Int. J. Elect. Power Energy Syst., vol. 26, no. 10, pp.
761–769, 2004.
[7] M. A. S. Masoum, A. Jafarian, M. Ladjevardi, E. F. Fuchs, and W. M.
Grady, “Fuzzy approach for optimal placement and sizing of capacitor
banks in the presence of harmonics,” IEEE Trans. Power Del., vol. 19, M. E. El-Hawary (LF’09) received the Bachelor’s
no. 2, pp. 822–829, Apr. 2004. degree in electrical engineering (Hons.) from the
[8] M. A. S. Masoum, M. Ladjevardi, A. Jafarian, and E. F. Fuchs, “Op- University of Alexandria, Egypt, in 1965, and
timal placement, replacement and sizing of capacitor banks in distorted the Ph.D. degree from the University of Alberta,
distribution networks by genetic algorithms,” IEEE Trans Power Del., Edmonton, 1972, where he was a Killam Memorial
vol. 19, no. 4, pp. 1794–1801, Oct. 2004. Fellow.
[9] X. Yu, X. Xiong, and Y. Wu, “A PSO-based approach to optimal capac- Currently, hs is Associate Dean of Engineering
itor placement with harmonic distortion consideration,” Elect. Power and has been Professor of Electrical and Computer
Syst. Res., vol. 71, no. 1, pp. 27–33, 2004. Engineering at Dalhousie University, Halifax, NS,
[10] T. M. Khalil, H. K. M. Youssef, and M. M. A. Aziz, “A binary par- Canada, since 1981. He served on faculty, and was
ticle swarm optimization for optimal placement and sizing of capacitor a chair of the Electrical Engineering Department
banks in radial distribution feeders with distorted substation voltages,” at Memorial University of Newfoundland for eight years. He was Associate
in Proc. AIML Int. Conf., Sharm El Sheikh, Egypt, Jun. 13–15, 2006, Professor of Electrical Engineering at the Federal University of Rio de Janeiro
pp. 129–135. for two years and was Instructor at the University of Alexandria. He pioneered
[11] J. H. Teng, “Network-topology-based three-phase load flow for distri- many computational and artificial intelligence solutions to problems in eco-
bution systems,” Proc. Natl. Sci. Counc. ROC (A), vol. 24, no. 4, pp. nomic/environmental operation of power systems. He has written ten textbooks
259–264, 2000. and monographs, and more than 120 refereed journal articles. He has consulted
[12] J. Teng and C. Chang, “Backward/forward sweep-based harmonic anal- and taught for more than 30 years.
ysis method for distribution systems,” IEEE Trans. Power Del., vol. 22, Dr. El-Hawary is a Fellow of the IEEE, the Engineering Institute of Canada,
no. 3, pp. 1665–1672, Jul. 2007. (EIC), and the Canadian Academy of Engineering (CAE).